bubbles_euler with num_fluids > 2 runs an unvalidated mixture closure

[sbryngelson]

case_validator.py accepts bubbles_euler with any num_fluids, but the ensemble-averaged model
MFC implements is derived for a single carrier liquid, and the code's own arms disagree above two
fluids. Found while reviewing #1762.

What the model says

The liquid phase closure is

$$\Gamma_l\,p_l = \frac{1}{1-\alpha}\left(E - \tfrac{1}{2}\rho|\mathbf{u}|^2\right) - \Pi_{\infty,l}$$

• the liquid's own Gamma_l and Pi_inf,l, undiluted, with the void entering only through the
  1/(1 - alpha) on the energy. s_compute_pressure and f_pressure implement exactly this.

The mixture pressure is p = (1 - alpha_b) p_l + alpha_b (...), with p_l a single carrier-phase
pressure. The formulation does not define several distinct liquid components alongside the bubbles.

What the code does

Under bubbles_euler the void fraction is the last advection slot
(eqn_idx%alf = eqn_idx%adv%end), so the material slots are 1..num_fluids-1. The mixture rule is:

num_fluids	mpp_lim	coefficients
<= 2	either	fluid 1's own, undiluted - matches the closure above
> 2	false	sum_{i=1}^{N-1} alpha_i gamma_i, which is the liquid mixture diluted by (1 - alpha)
> 2	true	sum_{i=1}^{N} alpha_i gamma_i, which includes the void slot

The first row is the published closure. The second is not obviously it: for
Gamma_l p_l + Pi_l = (E - rho|u|^2/2)/(1 - alpha) to hold with Gamma_l the true liquid-mixture
value, that sum would need dividing by (1 - alpha), since the material volume fractions sum to
1 - alpha rather than 1. The third row folds the void's gammas(num_fluids) into the mixture and
relies on mpp_lim having renormalized the fractions.

Why it has not surfaced

No test in the suite runs bubbles_euler with num_fluids > 1 at all; #1762 adds a two-fluid case,
so the > 2 rows remain uncovered.

Suggested resolution

Decide which of these is intended:

1. bubbles_euler supports one carrier liquid only, and case_validator.py prohibits
   num_fluids > 2 with a PHYSICS_DOCS entry. This matches the published model and is the smallest
   change.
2. A multi-component carrier is intended, in which case the num_fluids > 2 arm needs the
   (1 - alpha) normalization and a test that exercises it.

Until it is decided, num_fluids > 2 with bubbles_euler runs a closure nobody has validated.

[sbryngelson]

Resolved in #1762 by taking option 1: case_validator.py now refuses bubbles_euler with num_fluids > 2, with the closure recorded in its PHYSICS_DOCS entry. The two disagreeing arms are left in place but unreachable, so nothing changes for supported cases. If a multi-component carrier is wanted later, the (1 - alpha) normalization and a covering test are what that needs.